HareeshDaxton/Algerian-Forest-Fires-Regression-Analysis-Project

Predict forest fire occurrences using regression on Algerian weather data. Complete ML pipeline with EDA, Ridge Regression, Flask API, and AWS deployment. Built for real-world scalability and clarity. # Algerian Forest Fires – Regression Analysis Project This project presents a complete **Machine Learning Regression pipeline** to analyze and predict forest fire behavior using weather and fire index data from Algeria. From **data preprocessing and EDA** to **model training with Ridge Regression** and **deployment readiness via Flask and AWS Elastic Beanstalk**, this solution follows a professional and production-oriented workflow. --- ## 🗂️ Project Directory Structure ├── .ebextensions/ # AWS deployment configs (Elastic Beanstalk) ├── model/ │ ├── ridge.pkl # Serialized Ridge Regression model │ └── scaler.pkl # StandardScaler object used during training │ ├── NotBooks/ │ ├── raw_dataset/ │ │ └── Raw_Algerian_forest_fires_dataset_UPDATE.csv │ ├── cleaned_dataset/ │ │ └── Cleaned_Algerian_forest_fires_dataset.csv │ ├── ridge_EDA_FE.ipynb # Exploratory Data Analysis & Feature Engineering │ └── ridge_MAIN.ipynb # Model training and evaluation │ ├── application.py # Flask app for inference --- ## 📚 Dataset Overview - **Source**: UCI ML Repository – Algerian Forest Fires Dataset - **Context**: This dataset provides weather and fire index data collected from two Algerian regions during June–September. - **Files Included**: - Raw_Algerian_forest_fires_dataset_UPDATE.csv – Original dataset - Cleaned_Algerian_forest_fires_dataset.csv – Cleaned, preprocessed version used for modeling ### 🔑 Features Used - **Environmental**: Temperature, Relative Humidity (RH), Wind, Rain - **Fire Weather Indices**: FFMC, DMC, DC, ISI - **Target**: Classes → encoded to binary: 1 (fire), 0 (not fire) --- ## 🔄 End-to-End Workflow ### 📥 1. Data Loading & Cleaning - Combined datasets from both regions - Removed null values and fixed inconsistent entries - Cleaned labels and standardized formatting ### 📊 2. EDA & Feature Engineering - **Visualized** distributions, pairplots, and correlation matrices - **Detected a …